13:01:20 #startmeeting Network Health 2025-01-26 13:01:20 Meeting started Mon Jan 26 13:01:20 2026 UTC. The chair is hiro. Information about MeetBot at https://wiki.debian.org/MeetBot. 13:01:20 Useful Commands: #action #agreed #help #info #idea #link #topic. 13:01:25 #link https://pad.riseup.net/p/tor-nethealthteam-2026-keep 13:01:41 o/ 13:02:02 who would like to go first? 13:02:18 I just remember that today is a holiday in India and Sarthik is afk 13:03:40 I've been working on the anomalies and i've most of the code, but now investigating why the values aren't the same as in the original code 13:04:51 I am the same boat as you juga but with recreating the metrics from metrics.tpo into metricsdb 13:05:00 :) 13:05:01 so I feel your pain lol 13:05:05 lol 13:06:10 hiro: do you follow the steps on the reproducible metrics page? 13:06:17 yes 13:06:20 and they give you different results 13:06:24 ugh 13:06:37 is it for all metrics or just some of them where this is happening? 13:06:42 but there is deduplication in the code for when we have more than 1 reporting per day 13:06:56 relays I think I have it ok now 13:07:11 it is bridges that I am still overcounting 13:07:36 i recalled the numbers we had in victoriametrics were never matching because we did something different there 13:07:41 *rcall 13:07:45 *recall even 13:07:54 victoriametrics was doing its own dedup 13:07:54 okay, interesting 13:08:00 in metrics.tpo we have this https://gitlab.torproject.org/tpo/network-health/metrics/website/-/blob/master/src/main/java/org/torproject/metrics/stats/clients/Main.java?ref_type=heads#L382 13:08:42 sorry it's this function https://gitlab.torproject.org/tpo/network-health/metrics/website/-/blob/master/src/main/java/org/torproject/metrics/stats/clients/Main.java?ref_type=heads#L277 13:08:52 not the bytes parsing but the responses 13:09:30 hrm 13:11:12 I might have to go through again the sql too, which I haven't done yet 13:11:32 yeah 13:11:45 iirc some that is pretty wild :) 13:11:51 *some of 13:12:15 anyway that's my update basically 13:12:27 hope to get over with this soon and have some time to fix meronense 13:13:02 hiro: do we know why we miss consensuses on collector from time to time? 13:13:10 I have a doubt regarding nsa is it right time to ask here or should i pm hiro? 13:14:14 hey Rohithh you can ask me just after the sync on the Tor Network Health channel 13:14:59 GeKo (IRC): I am not sure about that, my take is that the service is busy doing something and it misses the window to request it because it always happens at the end of the month 13:15:00 i guess that's still collector#40057, okay 13:15:05 sure thanks 13:15:28 pretty often 4 or 5 am utc, hrm 13:15:59 hiro: no, it's not just at the end of the month 13:16:20 e.g. Expected in archive no later than2025-12-08T05:55Z 13:16:29 but anyway 13:17:07 as to my update: i looked a bit closer at the key network health indicators 13:17:11 uhm ok... I thought that was when it is updating the tarballs and it carries out for a week before and a week after the end of the month... but I'll check the alerts and see 13:17:37 and thought about how to further write the tor-anomaly document 13:18:02 hiro: i think i'll move a large part from the wiki into that document as well 13:18:17 taht is all the metrics/data we collect and where as an appendix ors omething 13:18:44 and then i'll have a small chapter about tor anomaly work and data/metrics we need for that 13:18:57 i plan to do all of that this week 13:19:07 that's it from my side 13:19:27 ok, when the metrics website would be closer to be published we could have a section there of all these docs 13:19:54 so interested people can easily find them all 13:19:59 worth thinking about, yes 13:20:30 on my side, I am still working on the bandwidth route and I expect it to be completed this week 13:20:45 yay nice thanks Rohithh 13:22:01 alright if everybody is groot we can end the meeting 13:22:10 i am 13:22:24 +1 13:22:24 I'm groot 13:22:32 #endmeeting